Zoffix \o/ 00:00
Now I have a fully-working test that covers the bug: gist.github.com/zoffixznet/84c4ba8...f68baea3ff 00:01
ZofBot: now I can be dangerous >:D
ZofBot Zoffix, He shut his eyes
Zoffix Yeah, there's MVM_file_handle_from_fd in moarvm 00:41
Got three test files failing on HEAD on macos: gist.github.com/zoffixznet/c5bba61...becaa28cdf 00:51
... 01:12
nm 01:13
MasterDuke Zoffix, timotimo: looking at github.com/rakudo/rakudo/commit/e4a5bb17c9, think it would make sense to special case dividing by (-)?1 in MVM_bigint_div and just return the numerator (possibly negated)? 01:16
Zoffix No idea. I'm only using it there 'cause it gives an uncached Int back 01:19
(nqp::add_I() for example doesn't) 01:20
MasterDuke heh, i wonder if that optimization would then negate the reason you're using it 01:21
because MVM_bigint_div always allocates a new mp_int for the result 01:22
and i was thinking about just returning what was passed in for the numerator if the denominator is 1
Zoffix Yeah, but that's not a reason to not do the optimization if it's needed :) 01:24
MasterDuke well, div_I is already faster than i expected, but if it could be made even faster... 01:25
Zoffix This is really pretty: i.imgur.com/kYSft5S.png would be cool to be able to show similar stuff in Rakudo (even if with some custom Exceptions:: handler module) 01:30
Well, I got a fix for $*IN.getc thing... it looks a bit nutty tho 01:54
And prolly makes things a bit slower on OSX
or macos whateve
Can an fd other than 0, 1, or 2 be some weird TTY thing? 02:07
Oh wait, yeah it can... My test file opens one..
BenGoldberg m: use NativeCall; sub isatty(int --> int) is native {...}; say isatty($_) for ^4; 02:19
camelia Potential difficulties:
In 'isatty' routine declaration - Not an accepted NativeCall type for parameter [1] : int
--> For Numerical type, use the appropriate int32/int64/num64...
at <tmp>:1
------> 3 sub isatty(int --> int) …
BenGoldberg m: use NativeCall; sub isatty(int32 --> int32) is native {...}; say isatty($_) for ^4;
camelia 0
0
0
0
Zoffix Looks like I killed stmuk's VM
Wonder if it's 'cause I spectested with my patch and the patch is crap :/ 02:20
it stopped producing output. I could still ssh into it, but no commands ran... Now even ssh doesn't respond. I guess my spectest ate all the resource 02:21
oh, nm, it cam eback
Zoffix tries spectesting again
Geth roast: bcf0f93354 | (Zoffix Znet)++ | packages/Test/Util.pm
Implement `run-with-tty` test routine

Similar to `is-run` except it does magiks to make standard handles be TTY so you could do TTY-sensitive tests.
02:30
roast/fix-macos-eof: 27833272b4 | (Zoffix Znet)++ | S16-io/eof.t
Test .eof on TTY handles works right

RT#132349: rt.perl.org/Ticket/Display.html?id=132349
02:35
synopsebot RT#132349 [open]: rt.perl.org/Ticket/Display.html?id=132349 [REGRESSION] $*IN.getc not blocking on macOS
Zoffix .tell lizmat are you able to try spectesting (or stresstesting) with this MoarVM branch? github.com/MoarVM/MoarVM/commit/24007720d5 It fixes $*IN.getc bug but I'm having trouble completing spectest on stmuk++'s VM; seems to run out of resources or something. Dunno if it's my change. If you do run it, can you note if there's any significant performance impact? 02:39
yoleaux Zoffix: I'll pass your message to lizmat.
Zoffix .tell AlexDaniel` for RT#132349 I got a solid test ( github.com/perl6/roast/commit/27833272b4 ) and questionable fix in a MoarVM branch ( github.com/MoarVM/MoarVM/commit/24007720d5 ). It fixes the bug, but maybe doing Bad Things™ or impacting performance. Dunno how good it is. I'm a n00b in this area. 02:41
yoleaux Zoffix: I'll pass your message to AlexDaniel`.
Zoffix drops to bed
timotimo mastertbh i have no idea why that's div_I and not add_I which was what i suggested 06:49
oh, now i see what's what
there should be something against mutated objects in the int cache anyway 06:52
[Tux] This is Rakudo version 2017.09-498-ge4a5bb17c built on MoarVM version 2017.09.1-614-g19523568 06:54
csv-ip5xs 1.163 - 1.179
test 11.844 - 11.908
test-t 3.196 - 3.205
csv-parser 12.000 - 12.223
timotimo right, the impl of div just doesn't look at the int cache 07:07
that's why you get a distinct object
AlexDaniel` . 07:08
yoleaux 02:41Z <Zoffix> AlexDaniel`: for RT#132349 I got a solid test ( github.com/perl6/roast/commit/27833272b4 ) and questionable fix in a MoarVM branch ( github.com/MoarVM/MoarVM/commit/24007720d5 ). It fixes the bug, but maybe doing Bad Things™ or impacting performance. Dunno how good it is. I'm a n00b in this area.
synopsebot RT#132349 [open]: rt.perl.org/Ticket/Display.html?id=132349 [REGRESSION] $*IN.getc not blocking on macOS
AlexDaniel` .tell Zoffix awesome. Thanks!
yoleaux AlexDaniel`: I'll pass your message to Zoffix.
AlexDaniel` lizmat: oh 07:09
lizmat: I forgot to generate it this week?
oops…
yep I did 07:11
timotimo ISTR jnthn discussing what to do about int cache and mixing in stuff 07:12
AlexDaniel` .tell lizmat Oops. I forgot to do it this week. Sorry. As for the script, I can share it if you want but I'm doing some weird stuff to get things working so maybe it'll be easier if I just give you a link every week. Tell me what you think
yoleaux AlexDaniel`: I'll pass your message to lizmat.
lizmat . 07:15
yoleaux 02:39Z <Zoffix> lizmat: are you able to try spectesting (or stresstesting) with this MoarVM branch? github.com/MoarVM/MoarVM/commit/24007720d5 It fixes $*IN.getc bug but I'm having trouble completing spectest on stmuk++'s VM; seems to run out of resources or something. Dunno if it's my change. If you do run it, can you note if there's any significant performance impact?
07:12Z <AlexDaniel`> lizmat: Oops. I forgot to do it this week. Sorry. As for the script, I can share it if you want but I'm doing some weird stuff to get things working so maybe it'll be easier if I just give you a link every week. Tell me what you think
lizmat AlexDaniel`: a link every week would be fine :-)
AlexDaniel` .at 2017-10-29 RT/GH weekly 07:17
yoleaux AlexDaniel`: Sorry, that command (.at) crashed.
lizmat Files=1229, Tests=75769, 317 wallclock secs (14.75 usr 5.13 sys + 2190.72 cusr 214.68 csys = 2425.28 CPU)
AlexDaniel` .in 5d RT/GH weekly
yoleaux AlexDaniel`: I'll remind you on 29 Oct 2017 07:17Z
AlexDaniel` .in 12d RT/GH weekly
yoleaux AlexDaniel`: I'll remind you on 5 Nov 2017 07:17Z
AlexDaniel` .in 19d RT/GH weekly
yoleaux AlexDaniel`: I'll remind you on 12 Nov 2017 07:17Z
AlexDaniel` .in 26d RT/GH weekly
yoleaux AlexDaniel`: I'll remind you on 19 Nov 2017 07:17Z
AlexDaniel` .in 33d RT/GH weekly 07:18
yoleaux AlexDaniel`: I'll remind you on 26 Nov 2017 07:18Z
AlexDaniel` mmm maybe all the way till the end of this year…
.in 40d RT/GH weekly
yoleaux AlexDaniel`: I'll remind you on 3 Dec 2017 07:18Z
AlexDaniel` .in 47d RT/GH weekly
yoleaux AlexDaniel`: I'll remind you on 10 Dec 2017 07:18Z
AlexDaniel` .in 54d RT/GH weekly
yoleaux AlexDaniel`: I'll remind you on 17 Dec 2017 07:18Z
AlexDaniel` .in 61d RT/GH weekly
yoleaux AlexDaniel`: I'll remind you on 24 Dec 2017 07:18Z
AlexDaniel` .in 68d RT/GH weekly 07:19
yoleaux AlexDaniel`: I'll remind you on 31 Dec 2017 07:19Z
AlexDaniel` too bad no yearly!
lizmat: I did start the script now again, but I think it'd be better if we show the report for the last two weeks next time 07:20
also, I need to add github support to the script
lizmat AlexDaniel`: yeah, next week a 2 week report would be fine :-) 07:21
AlexDaniel` OK
stmuk so no release until next month? 08:34
Zoffix: It's a physical laptop not a VM but its running the latest macOS which probably hasn't been tested much on that exact platform and I'm not sure how well it works with SSH (power suspend etc.?) 08:37
timotimo no release? why?
lizmat is still working on making issue 1220 go away 08:38
stmuk github.com/rakudo/rakudo/commit/76...299eca23a2 08:39
timotimo it has the 2017.10 already in the "finished" section 08:40
albeit with yesterday's date
stmuk oh yes
more coffee needed 08:41
stmuk tries the macOS moar branch
I should active the virtualsue bot to fix some clang warnings :) 08:45
lizmat afk for a few hours& 08:50
AlexDaniel` timotimo: oh well :) 09:17
the date in the release guide is always changed a bit ahead of time… even if by a few seconds :) 09:18
stmuk: pretty sure it won't take us a month to fix the remaining issues :) 09:19
stmuk Zoffix: I think there are performance issues with your fix-macos-eof (using clang on macOS 10.13) I'm seeing stage parse at 210 secs rather than 123 secs for moar-blead 09:38
its not as bad on linux/gcc 133 secs v 80 secs
AlexDaniel` hmm 09:52
stmuk I also saw t/spec/S06-signature/slurpy-params.t from 6.d-proposals 09:58
fail as well as some S17 (on macOS) and had to give up since the laptop was running too hot with fans going 09:59
Zoffix doubt my fix affects stage parse 10:16
yoleaux 07:08Z <AlexDaniel`> Zoffix: awesome. Thanks!
Zoffix gonna do what geekosaur++ suggested tonight; unless it gets merged/released before that 10:17
Also, my fix shouldn't affect linux at all :) 10:21
Zoffix spies a perl6 repo created for book leanpub.com/collab-dev-git 10:24
Oh, I guess there are many langs: github.com/collab-dev 10:25
AlexDaniel`: "yearly"? like instead of weekly? I think we have that in some form if someone does one: the Advent Posts. 10:41
We prolly should start soliciting authors 10:42
AlexDaniel` Zoffix: ah no, I meant that if I had done this before, then by the end of the year we could have had a yearly progress report for RT tickets
(like this: gist.github.com/AlexDaniel/0f20a50...d1d94002e2 ) 10:43
Zoffix ah 10:44
Zoffix will fire off a Call for Authors for advent in ~2h 10:51
I see last year we started on Nov 1st, and we were kinda in a bind to fill all the spots by the end... better start now.
AlexDaniel` Zoffix++ 10:52
Zoffix huh "You did it! Congratulations on submitting your four pull requests and contributing to open source projects this month" 11:00
I wasn't even trying... And I guess it does count PRs to brojects you got commits for, 'cause I don't remember submitting PRs anywhere else :/
*got commit bits for
ZofBot: plot twist: Zoffix sleep-walks and submits PRs at night without remembering anything the next day 11:01
ZofBot Zoffix, The room was a world, a pocket of the past where extinct animals could walk
stmuk stmuk-- # ah my mac eof branch moar build script had --no-optimize --debug and I forgot 11:38
Zoffix ah :) phew 11:39
stmuk and the slow test results are probably due to the aggressive TEST_JOBS=6 being better suited to beefy multi-core systems than ageing laptops 11:43
Geth rakudo/nom: 30462d766e | (Elizabeth Mattijsen)++ | src/core/ThreadPoolScheduler.pm
HLL-fix for issue 1202

  - more info: see github.com/rakudo/rakudo/issues/1202
  - changes all use of Lists for worker list to IterationBuffer
   - these have fewer concurrency issues
   - specifically, List.elems *could* change internal state of a List
  - this appears to make the sample code much more stable
... (5 more lines)
11:50
lizmat AlexDaniel`: ^^^ 11:51
I think I've done all I can do in this respect
AlexDaniel` great. I'll test it in a few moments 11:52
lizmat some extra datapoint: I can only get the code of issue 1202 to fail if I run something heavy like a spectest concurrently 11:57
if it runs by itself, it can run for minutes 11:59
stmuk t/spec/S17-supply/watch-path.t fails for me 12:13
Geth synopsebot: c93e528c35 | (Aleks-Daniel Jakimenko-Aleksejev)++ | 3 files
“GH #…” issues and PRs

Implemented similarly to RT plugin. There's some overlap in the code which can be refactored.
12:14
AlexDaniel` Zoffix: how to git pull & restart synopsebot? 12:15
Zoffix I was thinking of something slitly different that'd let us link to more than Rakudo's Issues. Syntax: RT#\d+ = RT ticket; R#\d+ = Rakudo's Issues, S#\d+ = Spec's Issues, M#\d+ = MoarVM's issues, N#\d+ = NQP's issues
AlexDaniel` hmm 12:16
Zoffix AlexDaniel`: ssh hack.p6c.org; sudo su -l synopsebot; cd synopsebot/; git pull; logout; sudo service synopsebot restart 12:17
I done it now
G#1202
GH#1202
synopsebot GH#1202 [open]: github.com/rakudo/rakudo/issues/1202 [annoying] Async qqx sometimes hangs or dies ( await (^5).map({start { say qqx{… …} } }) )
Zoffix neat
AlexDaniel` works for PRs also
GH #998 12:18
synopsebot GH#998 [open]: github.com/rakudo/rakudo/pull/998 [pending user] Implement `last` and `redo` for whenever blocks
stmuk and t/spec/S32-str/utf8-c8.t fails
AlexDaniel` Zoffix: ok, so someone suggested [severe] instead of [annoying], what do you think?
Geth synopsebot: 11d98e334e | (Zoffix Znet)++ (committed using GitHub Web editor) | README.md
document how to pull new changes and restart the bot
Zoffix AlexDaniel`: good
AlexDaniel` irclog.perlgeek.de/perl6-dev/2017-...i_15340708
ok then 12:19
done. Renamed. 12:22
Zoffix ZofBot: good 12:23
ZofBot Zoffix, For all he knew there might never have been any such law as the JUS PRIMAE NOCTIS, or any such creature as a capitalist, or any such garment as a top hat
Geth rakudo/nom: b19e352eca | (Elizabeth Mattijsen)++ | src/core/List.pm
Revert "Threadsafe List.reify-until-lazy a bit"

  - This reverts commit 50324bb004e120a0f82363bf8ff4222dc2cfa014.
  - No longer needed because of 30462d766e9216c9f5beed
  - Removes small overhead the commit caused
12:24
AlexDaniel` ahhh :) irclog.perlgeek.de/perl6/2017-10-24#i_15345901 12:27
pmurias should I file a ticket if I find a bug in rakudo and will fix it myself? 13:17
AlexDaniel` pmurias: IMO yes
DrForr I'd agree, someone else might find the same bug and duplicate your work. 13:18
AlexDaniel` pmurias: one of the reasons for “yes” is that people tend to find and fix issues but forget to write tests for them
also it's much better to have tests linked to some context
perlpilot AlexDaniel`: re synopsebot ... maybe someone should make a way for it to restart itself. :) 13:19
AlexDaniel` if you want to work on it yourself feel free to self-assign, that's possible on RT and GitHub
geez, was away for a lunch and my inbox is full of notifications 13:20
perlpilot: maybe. For example, camelia knows how to do that 13:21
lizmat: ok, without any load it hangs on my machine in ≈30 runs 13:24
sometimes more, sometimes less
but after your changes I've never seen any error
so it got better for sure 13:25
not much tho
pmurias should new tickets be opened on github? 13:29
Zoffix pmurias: whatever you feel more confortable. I think the plan is to let the best Issue tracker win :) 13:31
AlexDaniel` yeah
timotimo AlexDaniel`: could i have a feature that i give synopsebot a (abbreviated) commit hash and it'll figure out for me what repo it's from and generate a github link? 13:35
AlexDaniel` timotimo: yes
timotimo cool
AlexDaniel` although maybe that's better done in whateverable because it already tracks all repos
alright someone is saying that rakudo does not build for him. Great 13:36
Zoffix Two people actually. Another one filed on nqp repo last night 13:39
timotimo sure, put it on whateverable, that also makes sense
AlexDaniel` timotimo: there github.com/perl6/whateverable/issues/255 13:41
[Coke] btw, I found a "github issue mover" that we could use to move things from, say, docs repo to rakudo. 13:42
(since some doc tickets are really for things in perl6)
AlexDaniel` yeah I think this was mentioned previously 13:43
Zoffix: that one I'm not worried about. It builds just fine on debian unstable, there should be no problem on testing 13:44
Zoffix ZofBot: famous last words 13:45
ZofBot Zoffix, The economy of many countries was allowed to stagnate, land went out of cultivation, capital equipment was not added to, great blocks of the population were prevented from working and kept half alive by State charity
AlexDaniel` timotimo: what was the issue back then 3.5 years ago? irclog.perlgeek.de/moarvm/2014-05-11#i_8709129 13:47
:)
timotimo looks like i changed something in spesh optimization that caused code to misbehave once it got optimized 13:48
buggable New CPAN upload: JSON-Tiny-1.0.tar.gz by MORITZ cpan.metacpan.org/authors/id/M/MO/...1.0.tar.gz 14:22
AlexDaniel` ok so -2 build issues then 14:30
Zoffix New blog post: "Rakudo Perl 6 Advent Calendar 2017 Call for Authors": rakudo.party/post/Rakudo-Perl-6-Ad...or-Authors 14:42
DrForr RT'd. 14:45
Zoffix stmuk: is your macos lappy available for use ATM? Was gonna try the isatty now instead of in 8hr 15:05
guess I can try it without `#if defined` things on my 24-core Linux to see if there's any perf impact 15:06
ugexe travis-osx-reverse-shell-serverable: 15:31
github.com/fniephaus/travis-remote-shell >:-) 15:33
that one doesnt work on osx builds i guess, but there are others 15:34
Zoffix ZOFFLOP: t/spec/S11-modules/nested.t 15:39
HEAD: Stage parse 79s; stresstest: 157s; spectest: 112s; bench: 6.560s 15:42
"bench" being this: time ./perl6 -e '(my $p = $*TMPDIR.add: "z42") andthen .spurt: "meows"; for ^50000 { given $p.open { .eof; .slurp; .eof; .close } }'
.oO( useless use of andthen )
Slightly too inflated bench; on 472 I get 6.209s 15:48
m: $*PERL.compiler.version.say
camelia v2017.09.500.gb.19.e.352.ec
Zoffix ZOFFLOP: t/spec/S17-lowlevel/semaphore.t 15:57
(that's a new one :o)
Segfaulted during the run
HEAD: Stage parse 79s; stresstest: 157s; spectest: 112s; bench: 6.209s-6.560s 16:03
EOF fix: Stage parse 74s; stresstest: 153s; spectest: 112s; bench: 6.236s-6.406s
so perf impact is within noise
rofl. So it begins, the AI is attacking :) www.theverge.com/us-world/2017/10/...od-morning 16:09
ZofBot: you'll spare me, won't you?
ZofBot Zoffix, 'It's nothing
Zoffix "translation systems made an error" or the AI wanted us to think it made an error :P 16:11
ilmari google translates the phrase as "become them" 16:12
Zoffix .tell AlexDaniel` I updated the $*IN.getc PR with the new suggestions. I don't got a macos to test this on, but when I made the patch applicable to Linux and stresstested without issues and no noticable perf impact. If someone tests this on macos and it fixes the bug, I think it's good to merge: github.com/MoarVM/MoarVM/pull/731 16:15
yoleaux Zoffix: I'll pass your message to AlexDaniel`.
Zoffix & for a few hours
AlexDaniel` cool 16:16
yoleaux 16:15Z <Zoffix> AlexDaniel`: I updated the $*IN.getc PR with the new suggestions. I don't got a macos to test this on, but when I made the patch applicable to Linux and stresstested without issues and no noticable perf impact. If someone tests this on macos and it fixes the bug, I think it's good to merge: github.com/MoarVM/MoarVM/pull/731
[Coke] testing now. 16:30
(er, doing the build now.)
Zoffix is about to commit slight restructure tho :)
[Coke] ok. killing the build, let me know when to retry 16:34
Zoffix [Coke]: ok, pushed my latest work 16:47
github.com/MoarVM/MoarVM/commit/fbd7dc8fa5
[Coke] same branch? restarting.. 16:50
... did you force push?
got a merge conflict having done nothing locally. 16:51
repulling, this will add many minutesover my slow connection. :|
actually, it's just way too slow. killing the build, will kick it off again when I'm back in the office. 16:53
AlexDaniel` so the thing with WebSocket module, it never gets anything here: github.com/tokuhirom/p6-WebSocket/...6W.pm6#L57
but at the same time I don't really know if anything is sent, hmm 16:54
Zoffix [Coke]: yeah, forced pushed
AlexDaniel` well maybe I can see in wireshark or something
[Coke] samcv: found the bug that hangs the async highlights on os x. 16:56
samcv oh?
awesome!
[Coke] m: say $*TMPDIR; 16:57
camelia "/tmp".IO
[Coke] ^^ on a mac, that path ends with a /
Zoffix :(
[Coke] so the tmp files we make there have a // in the middle; coffee is normalizing it, so the filenames never match.
AlexDaniel` yeah, stuff is sent but is never received 16:58
Zoffix m: '/tmp'.IO.say
camelia "/tmp".IO
Zoffix Does that give '/tmp/' on MacOS? 16:59
[Coke] no. 17:00
AlexDaniel` greppable6: HTTP::Server::Tiny
[Coke] I assume it's the construction of TMPDIR
greppable6 AlexDaniel`, gist.github.com/5f19c8884de0e170a3...b9c96032b7
[Coke] ah, it's that way in ENV.
so perl6 might want to standardize that 17:01
in the meantime, I'll just strip off any trailing /
(in docs)
Zoffix m: dd %*ENV<TMPDIR>
camelia Any <element> = Any
Zoffix Damn. All these discussions make me want to have a mac to work with... but I feel like Apple would've gotten away with its con if it got $x amount of money outta me just cause it's refusing to make MacOS easily VMable 17:02
[Coke] samcv: if I fix this, is there any reason to ever NOT use the async version? 17:03
I can rip out the command line arg and the ability to force single threaded also.
Zoffix That macincloud thing would've been alright if they didn't expire your credits after 60days 17:05
oh wait, it expires "if you do not log in" ok, maybe it's fine then 17:06
"Enable Remote Build Port (SSH) - $5/month per server" -_- 17:07
[Coke] samcv: will remove force-proc-async and the macosx specific code in MAIN; I'll leave the ability to use the non-async one, but IMO we should plan to rip it out. (and perhaps I will as I do more here.) 17:10
Zoffix :S CAD$3,699.99 for a decent macbook with a tiny 15.4" screen... the macincloud looking better by comparison :P 17:11
AlexDaniel` does anybody have any thoughts on github.com/tokuhirom/p6-WebSocket/issues/15 ? 17:18
especially github.com/tokuhirom/p6-WebSocket/...-339065577
it feels like a real sched bug 17:19
Zoffix #1111 17:33
Fixes #1111
synopsebot GH#1111 [closed]: github.com/rakudo/rakudo/pull/1111 begin_time_lexical_fixup for constants
Zoffix The bot's buggy. It just assumes rakudo issues
Like here: irclog.perlgeek.de/perl6/2017-10-24#i_15347647
AlexDaniel` #132353 17:34
ugh
blah #132353 17:36
synopsebot RT#132353 [new]: rt.perl.org/Ticket/Display.html?id=132353 [LTA] error with using meta assign ops with bound SetHash
AlexDaniel` blah #68320 17:37
synopsebot RT#68320 [open]: rt.perl.org/Ticket/Display.html?id=68320 [TODO] $!.pending
AlexDaniel` Zoffix: I guess it shouldn't react to things that have no prefix
Zoffix Yeah, probably
And make different prefexes for different locations. 17:38
Well, looked around for a MacOS VM. Don't see anything I'd be willing to pay for. The cheapest I saw was like CAD$46 for 30 hours and they round it to the hour and require you to log in at least once (and thus spend at least an hour) every 60 days or all your hours expire. 17:39
I'm too cheap to pay that to debug MacOS-specific issues.
[Coke] Zoffix: build fails. 17:41
gist.github.com/coke/9939bc56dfd59...af016c667a
I wonder if TPF has a mac contact. 17:42
perl Configure.pl --gen-moar=fix-macos-eof --gen-nqp=master && make -j install # build command
Zoffix dam 17:44
Zoffix tries a linux build with mac defines removed
[Coke]: do you still have previous output available? What's the output before the failures? 17:48
it built on linux... 17:49
stmuk Zoffix: do you need access to my mac? 17:50
Zoffix stmuk: that would be great. yeah 17:51
[Coke]: nm, managed to repro it on linux if I just use the `isatty` test. 17:52
Ah lol, Perlism in C code: missing `return` statement 17:58
stmuk Zoffix: ok its charging now and should be available although it might be better to let it charge for an hour or two before heating it up :) 18:00
Zoffix stmuk++ thanks
stmuk I think its possible to run macOS in virtualbox and even on PC hardware directly 18:04
Zoffix I found this article: www.pcsteps.com/2157-mac-os-x-virt...re-player/ Might try it; though comments suggest it's not as easy as pie 18:05
stmuk or watch www.puredarwin.org/ for progress .. last time I looked it only booted in single user mode and the guy on IRC tried to get me to join up and help them fix it :)
Zoffix :) 18:06
ZOFFLOP: t/spec/S11-modules/nested.t t 18:14
Yup. isatty() thing fixes the bug too. 18:28
Zoffix tries a TEST_JOBS=2 spectest 18:29
Geth synopsebot: 1db7e3e8a5 | (Aleks-Daniel Jakimenko-Aleksejev)++ | 2 files
More prefixes, stricter regex and better handling

A bit crude but it works.
IDs without prefixes are no longer accepted. Now you can mention both RT and GH tickets in the same message. Adds prefixes for other repos, adjust to your liking.
AlexDaniel` Zoffix: the code kinda sucks but it works ↑
example: NQP#100 MOAR#100 GH#1202 RT#132353 RAKUDO#1203 SPEC#100 18:31
synopsebot RT#132353 [new]: rt.perl.org/Ticket/Display.html?id=132353 [LTA] error with using meta assign ops with bound SetHash
Zoffix R##1203
R#1203
RAKUDO is too much to type 18:32
AlexDaniel` R is a bit too little :)
Zoffix Why?
AlexDaniel` well, we have R and RT
Zoffix Support both "R" and "RAKUDO"
So?
RT# and R# clearly distinct
Geth synopsebot: 5d3dc13d53 | (Aleks-Daniel Jakimenko-Aleksejev)++ | lib/SB/Plugin/GitHub.pm6
Add R# prefix

  irclog.perlgeek.de/perl6-dev/2017-...i_15347995
18:34
Zoffix Well, I meant, M, S, N, R :P
and D
Well, I can add myself, if you want 18:35
Geth synopsebot: 565bd86360 | (Aleks-Daniel Jakimenko-Aleksejev)++ | lib/SB/Plugin/GitHub.pm6
More confusing prefixes

  irclog.perlgeek.de/perl6-dev/2017-...i_15348014
AlexDaniel` I guess it doesn't really matter if these prefixes are meaningful or not 18:36
Zoffix The docs are missing :}
DOCS#42 / D#42 18:37
Geth synopsebot: 5d69016616 | (Aleks-Daniel Jakimenko-Aleksejev)++ | lib/SB/Plugin/GitHub.pm6
Add DOCS# prefix
18:38
AlexDaniel` I wonder if all that is needed, but we'll see how it goes 18:39
Zoffix N#100 M#100 R#100 RT#132353 S#100 18:41
synopsebot RT#132353 [new]: rt.perl.org/Ticket/Display.html?id=132353 [LTA] error with using meta assign ops with bound SetHash
N#100 [closed]: github.com/perl6/nqp/issues/100 Forcing 64bit mode on OS X causes NativeCall to fail on 32bit machines
M#100 [closed]: github.com/MoarVM/MoarVM/issues/100 spesh bug with perl6-lwp-simple get-w3-latin1-utf8.t
R#100 [closed]: github.com/rakudo/rakudo/pull/100 Typename suggestions: in signatures, declarations, return types, ...
S#100 [closed]: github.com/perl6/roast/pull/100 Test for require file where symbol isn't in module
Zoffix neat
D#100
synopsebot D#100 [closed]: github.com/perl6/doc/pull/100 IO::Path .extension, Proc[::Async] :$cwd param
Geth synopsebot: 30ed232a8c | (Zoffix Znet)++ (committed using GitHub Web editor) | README.md
spread out commands on new lines

  'cause they can't be copy-pasted in one line
18:42
[Coke] Zoffix: ssh -t hack.p6c.org 'long list of commands that include sudo' 18:47
then you could cut and paste it, no?
Zoffix [Coke]: no idea. I'm a ssh n00b 18:48
one of commands is switching to synopsebot user to pull
This really should be a bot command
[Coke] you need the -t for the password prompt to work, but yes, that should do it.
Zoffix Weird that rakudo#44 doesn't work. Even tho there's an :i switch
AlexDaniel` ouch 18:49
hmm rakudo#100 18:50
hmm rakudo#1202
Zoffix m: my %URLS = :42FOO; say "foo#42" ~~ m:ex/:i « (@(%URLS.keys)) \s* '#' \s* (<[0..9]>**{2..6}) »/ 18:51
camelia (「foo#42」
0 => 「foo」
1 => 「42」)
Geth synopsebot: f57f96bcdd | (Zoffix Znet)++ (committed using GitHub Web editor) | lib/SB/Plugin/GitHub.pm6
normalize prefix
Zoffix rakudo#100 18:52
synopsebot RAKUDO#100 [closed]: github.com/rakudo/rakudo/pull/100 Typename suggestions: in signatures, declarations, return types, ...
Zoffix k; it was just not finding proper URL to visit
AlexDaniel` but why?
ah
Zoffix weird spectest looked "hung" on macos with my fix I pressed enter a few times and it looked to continue... Wonder if it's sleep mode, glitch due to this being a connection through like 4 different boxes, or if it's waiting for ENTER to be pressed :/ 18:56
don't look like any tests that touch filehandles, so I'm leaning towards 4-boxes option 18:57
stmuk I was having more success on that system with TEST_JOBS=3 19:00
Zoffix using 2 ATM 19:02
[Coke] Zoffix: one failure on that branch with spectest: 19:40
t/spec/S32-num/int.t
Zoffix [Coke]: which tests? 19:41
[Coke] not ok 1 - returns a new object (not a cached constant)
which I think is also you, but not this change.
Zoffix [Coke]: is it possible your rakudo is older than 1 day? That fix went in last night
like 24h ago
[Coke] ... oh. the new test isn't on roast master, so I didn't run that one.
Zoffix ./perl6 -e '$*IN.eof.say; $*IN.slurp.say; $*IN.eof.say' 19:42
[Coke] ah, yes, running a day old rakudo build.
Zoffix Run that, type something, it should print: "False\nyour something you typed\nTrue"
(taht's what the test in the branch does) 19:43
[Coke] yup
Zoffix \o/
OK. On st muk's laptop, I get the same failures with my patch as with HEAD yesterday: gist.github.com/zoffixznet/c5bba61...becaa28cdf 19:45
So imma wait for travis to clear that MoarVM PR and then merge and bump
um, looks like the new test hangs on macos, tho it's most likely the `script` args. Gonna poke at it in ~2h 19:53
Zoffix &
AlexDaniel` fwiw the release is not gonna make it today I think 20:00
stmuk today which TZ? 20:16
AlexDaniel` UTC-12⌁UTC+14 20:26
Zoffix Also why? As in, is there something I should look at this evening?
AlexDaniel` Zoffix: maybe, one sec 20:27
Zoffix: github.com/tokuhirom/p6-WebSocket/...-339120879 20:28
Zoffix k 20:30
AlexDaniel` I tried with RAKUDO_SCHEDULER_DEBUG=1 and RAKUDO_SCHEDULER_DEBUG_STATUS=1 but it didn't really give me any clues 20:31
if I recall correctly jnthn may be back late Wed, and I'll be available again around the same time 20:35
stmuk: any reason why you want the release to happen asap? :) I mean, we could've released it on Saturday, but then wasn't it you who was struggling with patches for outstanding bugs in releases? 20:41
also, last two times we were a bit too fast, so there were some point releases. So IMO being a few days off is better than having a sloppy release. 20:43
stmuk AlexDaniel`: I'd rather the release be solid than fast 20:44
AlexDaniel` .tell jnthn if it happens that you come back before we figure it out, here is a thing to look at github.com/tokuhirom/p6-WebSocket/...-339120879 (the release was delayed for other reasons anyway, so would be great to fix this thing also) 20:48
yoleaux AlexDaniel`: I'll pass your message to jnthn.
AlexDaniel` .tell samcv in case you miss the discussion: irclog.perlgeek.de/perl6-dev/2017-...i_15348425 20:49
yoleaux AlexDaniel`: I'll pass your message to samcv.
samcv hello
yoleaux 20:49Z <AlexDaniel`> samcv: in case you miss the discussion: irclog.perlgeek.de/perl6-dev/2017-...i_15348425
samcv thank you :) 20:50
AlexDaniel` that assuming that github.com/MoarVM/MoarVM/pull/731 is finally nailed :)
AlexDaniel` calls it a day 20:53
have a nice * everyone
samcv have a nice * too
[Coke] ~~ 20:54
samcv: any opinion on removing the non-proc::async path for doc highlighting? 20:58
samcv i'm not opposed to it as long as it works on mac and linux and maybe windows 20:59
[Coke] we're definitely using the async path on windows, it works on mac now that I've fixed it.
I don't think anyone is using windows to build the docs atm, but could be wrong. 21:00
er. that first windows should be linux
Zoffix Weird. This sits and waits on MacOS and I think it makes sense (.Bool `await`s the proc while the proc sits waiting for stdin things to happen), but on Linux it doesn't do that and the test completes : gist.github.com/zoffixznet/62f1f45...2b745466f4 21:44
Completes, but with output that's missing .slurp: # Got STDOUT: "False\n\nTrue\n" 21:46
Like looks like the awaition just said fuggit about waiting for STDIN to close and just moved on with its life
Geth roast/fix-macos-eof: e330e9a6b9 | (Zoffix Znet)++ | S16-io/eof.t
Fix test's failure mode

On failure, it diag()s the received output, but diag returns True, causing the failed test to succeed
21:48
roast/fix-macos-eof: fd9d9a2117 | (Zoffix Znet)++ | packages/Test/Util.pm
Reword proc launch

We can't .Bool a Proc as that'll cause its `await`ion before we sent stuff to STDIN
21:49
Zoffix booo... this year's Hacktoberfest shirts are uggo 22:05
they come in burn-out-your-retina blue with weird silver design i.imgur.com/0ZN7ml9.png 22:06
(what's with the wheat designs?)
samcv wow that's bad Zoffix 22:08
suddenly i don't want one anymore
Zoffix :P
samcv maybe that's their secret plan
Zoffix ZOFFLOP: t/spec/S11-modules/nested.t 22:11
t/spec/S17-channel/stress.t took ages again: 278s total stresstest time 22:12
Geth nqp: 8fa082b269 | (Zoffix Znet)++ | tools/build/MOAR_REVISION
Bump MoarVM
22:13
¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...1-g27f9134
rakudo/nom: eb1febd565 | (Zoffix Znet)++ | tools/build/NQP_REVISION
Bump NQP
¦ rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....6-g8fa082b
roast: zoffixznet++ created pull request #349:
Fix macos eof
roast: cb7ec603cc | (Zoffix Znet)++ (committed using GitHub Web editor) | 2 files
Fix macos eof (#349)

  * Test .eof on TTY handles works right
RT#132349: rt.perl.org/Ticket/Display.html?id=132349
  * Fix test's failure mode
... (8 more lines)
synopsebot RT#132349 [open]: rt.perl.org/Ticket/Display.html?id=132349 [REGRESSION] $*IN.getc not blocking on macOS
Zoffix ZOFFLOP: t/spec/S11-modules/require.t 22:21
ZOFVM: Files=1283, Tests=152772, 155 wallclock secs (20.96 usr 3.16 sys + 3349.06 cusr 164.05 csys = 3537.23 CPU) 22:22
Made a bit of a mod to my shell prefix line. Made it possible to label terminals so I know wtf each one is about when I got a bunch open: i.imgur.com/ncdMg7M.png 23:07
Relevant part of ~/.bashrc: gist.github.com/zoffixznet/1724806...d3cc5affc1 23:08
.oO( include perl6 version... )
23:09
neat: i.imgur.com/1iqQc4w.png code: gist.github.com/zoffixznet/f2a1b68...ce428ba529 23:18
our start time is noticable tho :P 23:19
timotimo any reason you're not using perl6 --version instead? 23:58
that's a whole lot faster 23:59